ARTICLE IN PRESS

Energy Policy 34 (2006) 72–87 www.elsevier.com/locate/enpol

CO2 emission trading within the European Union and Annex B countries: the cement industry case La´szlo´ Szabo´, Ignacio Hidalgo, Juan Carlos Ciscar, Antonio Soria Joint Research Centre, Institute for Prospective Technological Studies (IPTS), European Commission EXPO Building, Isla de la Cartuja, E-41092 Seville, Spain Available online 31 July 2004

Abstract The cement industry is the third largest carbon emitting industrial sector in the EU. This article presents the foreseeable technological evolution of the cement industry under business as usual circumstances, and examines the effects on the sector of carbon trading. For those purposes a global dynamic simulation model of the cement industry (CEMSIM) has been developed. The model is composed of a series of interconnected modules on cement consumption and production, international trade and capacity planning. This study quantifies the benefits achieved from emission trading in different markets (EU15, EU27 and Annex B), derived both from the revenues of emission trading, and from the lower compliance costs. The magnitude of the potential carbon leakage effect is also assessed. r 2004 Elsevier Ltd. All rights reserved. Keywords: Cement industry; CO2; Emission trading; Simulation model

1. Introduction Emission trading is one of the international flexible mechanisms of the Kyoto Protocol on climate change to reduce greenhouse gas emissions in a cost-efficient way. Overall compliance costs of meeting a given carbon reduction target are lower under an international emission market than in the case of segmented or isolated markets. The European Union is to set up a carbon emission market for the power and energyintensive industrial sectors (such as steel, cement, and pulp and paper) in 2005,1 i.e. 3 years before the start of the commitment period of the Kyoto Protocol (2008–2012). This will be the first multi-country carbon emission market in the world.

Corresponding author. Tel.: +34-954-488-236; fax: +34-954-488279. E-mail address: [email protected] (L. Szabo´). 1 Directive 2003/87/EC, 13 October 2003 (European Commission, 2003a). See also European Commission (2000b, 2003b).

0301-4215/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.enpol.2004.06.003

This article presents a global simulation model (CEMSIM—Cement Simulation Model)2 to quantitatively analyse the future development of the cement sector and the impacts of different carbon trading schemes on it. The cement industry is one of the largest energy consuming sectors. Indeed, the building material sector—whose emissions are dominated by the cement production—is the third largest CO2 emitting industrial sector world-wide and in the European Union.3 According to the importance of the cement industry many studies in the literature have dealt with its future 2 The cement industry model has been prepared with the aim to complete the POLES energy model (Prospective Outlook for the Longterm Energy System, (European Commission, 1996)), which is a partial equilibrium model of the world energy system. Besides the comprehensive energy supply and transformation modules, the POLES model has a module on the large energy-intensive industries (such as iron and steel, etc.). These modules are under revision in order to attain more detail on the carbon emission trade amongst the sectors covered by the EU proposal. The steel module has been the first to be reviewed (Hidalgo et al., 2003). 3 According to the 1990 emission levels (Capros et al., 2001).

ARTICLE IN PRESS L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

prospects. The IEA Greenhouse Gas R & D Programme study (see International Energy Agency (IEA), 1999) has global coverage and assesses the development of the sector in the 2020 horizon. Its main focus is on the CO2 reduction potential of the sector, examining the energy efficiency improvement options. Its main finding is that under a business as usual scenario global emissions from the industry almost double in the 1995–2020 period. They conclude that assuming considerable technological improvements, the increase in carbon emissions could be limited to 35%. Two other references of the literature are countryspecific, focusing on large cement consuming regions of the world. Worrell et al. (2000) describe the energy efficiency improvement potentials in the US cement industry, based on a detailed national technology database. These authors set up an energy conservation curve, and estimate an 11% energy saving potential for the country. Liu et al. (1995) follow a different approach for China and estimate a 30% energy intensity improvement potential. Their results are based on three different scenarios concerning the technology and efficiency of the new plants. Rotman and Kelmanzky (1997) estimated a function of cement demand in Argentina using regression analysis. They conclude that income, the construction price index, and cement consumption lagged one period are significant variables. According to their findings the income and lagged consumption are the most important variables, while the price of construction is less influential. The CEMSIM model is composed of a series of interconnected modules on cement consumption and production, international trade and capacity planning. Those variables are calculated using behavioural equations. The model covers 51 regions of the world, including all 15 EU countries,4 which enables to simulate the interactions of the different national emission trading markets. In the present modelling exercise the effects of three alternative policy scenarios are examined, differing in market coverage and equilibrium carbon prices. The dynamic recursive simulation model is taking into account rich technological details. Several technology options are considered concerning the future development of the industry. Feasible retrofitting options and emerging technologies are included in the technology database, which contains detailed information on investment, variable and retrofitting costs. This database, and in particular the retrofitting options, provide with the foundation of the marginal abatement cost curves for the industry, which is the core element of the emission trading simulations. Emission reductions are 4 Luxembourg and Belgium are merged together, while all other EU countries are modelled individually.

73

not based on static abatement cost curves, as in many similar models, but on a dynamic system where the interaction among price variations, international trade, technology options and economic logic drive the actions of the market participants. This article has the following structure. Section 2 gives a short description of the cement industry and introduces the retrofitting options for the sector. Section 3 describes the main characteristics of the CEMSIM model. Section 4 presents the results of the reference and the emission trading scenarios, with a detailed sensitivity analysis. Finally, Section 5 concludes.

2. The cement industry overview Cement is a basic material for building and civil engineering construction. Since cement production and consumption are directly related to almost all economic activities, they closely follow economic trends. The world production of cement grew from 594 million tonnes in 1970 to over 1500 million tonnes by 1997, showing an uninterrupted and steady growing trend (CEMBUREAU, 1999b, 2002). Table 1 presents the shares of cement production and consumption for different world regions in 1997. The European Union has a production and consumption share of around 11%. China alone accounts for one third of world production. The difference between the global consumption and production data indicates the size of net foreign trade, which accounts for 7% of cement consumption. It is noticeable that in all regions cement demand is mostly satisfied domestically. Transportation costs are relatively high compared to the cement price, which makes the transport of the cement expensive on long ranges. Furthermore, the raw material of cement production can be found everywhere, reducing the role of international trade. Cement is hardly transported for more than 150 km inland. However bulk transportation by sea is economically feasible, and Table 1 Share of cement production and consumption in 1997 Regions

Production (%)

Consumption (%)

Per capita consumption (kg)

EU15 Rest of Europe FSU Africa North America Latin America China India Rest of Asia OECD Pacific

11.3 5.5 2.6 4.2 6.1 7.5 33.2 5.3 18.1 6.5

10.9 4.8 2.8 4.6 6.8 7.2 32.9 5.1 19.2 5.7

446 363 137 91 348 271 404 82 270 568

ARTICLE IN PRESS 74

L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

actually plays a growing role. Data indicate that North America and Rest of Asia are the biggest importers, while the European Union is a net exporter of cement, together with the OECD Pacific (mainly Japan). Table 1 also presents the regional per capita cement consumption. The OECD Pacific (568 kg) and the European Union (446 kg) regions have the highest levels. Developing regions have much smaller per capita consumption, in the range of 91–270 kg. On the contrary China has an exceptionally high value amongst them (404 kg). Global cement production has had an average growth rate of 3.6% between 1970 and 1997. A wide range of growth rates can be found amongst the regions. The EU15 has been relatively stable over this period, with an average decline of
0.1%. Eastern Europe and the Former Soviet Union had two distinct phases in cement production. Till 1988 cement production grew at a relatively modest rate of 2.3%. During the economic crisis of transformation the production level dropped by more than 12%. The African region had an average increase of 4.5% over the period, while Asia (without China and India) had a growth rate of 7.5%. China experienced an average annual growth of 12% in this period, while India grew at an average rate of nearly 7%. These two countries are responsible for most of the world cement production and consumption expansion. North America had a stable 0.5% average growth in cement production, while Latin America had growth rates ranging between 5% and 12% in some periods, but subsequent crises in large producing countries (e.g. Brazil and Mexico) reduced its average growth in the overall period (CEMBUREAU, 1999b). 2.1. Energy use in cement manufacturing The cement industry is an energy-intensive industry with energy typically accounting for 30–40% of production costs. The energy consumption is estimated at about 2% of world total, and almost 5% of industry total. The fuel mix in the industry is usually carbonintensive, and calcinating process itself produces high amounts of CO2, which means that the cement industry contributes by 5% to global CO2 emissions (International Energy Agency (IEA), 1999). The role of the cement industry as a source of pollution will be even higher in the future, taking into account the prospects of very fast growth rates of large developing regions, such as China, India and South East Asia. In a sector characterised by such market expansion, energy efficiency improvements become crucial in the future development of the sector from an environmental point of view. In order to capture these improvement options, the CEMSIM model clusters the numerous available technology options into two main categories. One of them relates to energy efficiency improvements

Table 2 Average energy consumption of the dry and wet processes (GJ/t) Heat requirement

Dry process

Wet process

Chemical reactions Evaporation of water Heat lost Total

1.76 0 1.4 3.2

1.76 2.4 1.7 5.8

to be attained by building new capacities, while the other consists of retrofitting existing plants. Hence to fully understand the modelling approach a brief technological description follows. The four distinguished production steps of cement manufacturing are the following: 1. Mining of raw materials. 2. Preparation of raw materials: homogenising and grinding of raw material. 3. Burning of raw material to cement clinker. 4. Finish grinding of clinker and mixing with additives. The preparation of raw materials and finish grinding steps (stages 2 and 4) consume mainly electricity. During the clinker burning step (stage 3) prepared raw materials are burned at high temperatures (with a peak of over 1450 1C), first calcinating the materials. With the rising temperature the feed material starts to form clinker. At temperature above 900 1C, the calcinating takes place: CaCO3 ! CaO þ CO2 : At this step the generated CO2 leaves the material, the raw material losing more than one third of its original weight. This means that at the same time there is an additional source of CO2 emissions from the process itself. This is not CO2 from fuel burning, and it should be accounted for as anthropogenic CO2 emission. In the clinkerisation process, which takes place at 1300–1450 1C, parts of the material become liquid, forming nodules known as clinker. The energy consumption for the two basic process types is summarised in Table 2 (European Commission, 2000a; International Energy Agency (IEA), 1999). The average energy consumption of the wet process almost doubles that of the dry process, because the evaporation of water from the feed material requires substantial energy. 2.2. Technology selection Monitoring an appropriate technology set is fundamental to address the environmental performance of the industry. It should include the relevant technologies of the market in the time horizon of the model (till 2030). As the aim of the model is to analyse the energy consumption and carbon emission of the sector, the

ARTICLE IN PRESS L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

decisive factor has been the energy required by the different technologies. Seven technologies have been chosen to be included in the model according to the clinkerisation process (kiln type): 1. 2. 3. 4. 5. 6. 7.

wet rotary kiln semi-wet rotary kiln semi-dry rotary kiln dry long rotary kiln dry rotary kiln with preheater dry rotary kiln with preheater and pre-calciner shaft kiln.

The first six technologies have a common feature: their processes apply vertical rotary kilns. However, they are different in the water content of the feed material, which influences the energy requirements of the process. The more water the feeds contain, the more homogeneous the meal for the kiln is, resulting in a more homogenous clinker as well. At the same time the energy consumption is higher, as the wet material needs more energy to dry. Another important difference is the existence of a preheater or pre-calciner unit. The preheater kilns usually have multi-staged (4–6) cyclone preheaters. The raw material travels through the cyclones, with each subsequent cyclone having higher temperature. The energy consumption of kilns with suspension preheaters is much smaller, than the previous categories. They consume between 2.9 and 3.5 GJ/t. As one part of the calcination already takes place in the preheater, it has been possible to reduce the length (and energy consumption) of the kiln significantly, which has considerably reduced the energy consumption of the whole process. In case of pre-calciner kilns an extra combustion chamber is installed between the preheater and the kiln. This pre-calciner chamber consumes 60% of the fuel used in the kiln, and 80–90% of the calcination takes place there. There are many advantages of the precalcinating process. Firstly, it further decreases energy consumption by 8–11%. Secondly combustion in this chamber is at lower temperature than in the kiln, so lower grade fuel can be used (e.g. wastes, waste fuel), which diminishes the NOx emissions as well. New precalciner kilns could have capacities increased up to 12,000 t/day, which further increase their efficiencies.5 Shaft kilns are vertical installations, where raw materials are travelling from the top to the bottom by gravity. Raw material is mixed with the fuel and fed from the top, while air is blown from the bottom. This technology has the same processes as the others, namely calcination, clinkerisation and cooling. There are certain disadvantages of the technology. Clinker quality is 5

In Europe the usual kiln size is between three and five thousand t/ day.

75

highly dependent on the homogenisation of pellets and fuel, as well as on the air supply. Inadequate air supply or uneven air distribution makes combustion incomplete, resulting in low quality clinker and high CO and VOC emissions. In theory, shaft kilns can almost reach the energy efficiency of the rotary kilns, but in practice their energy consumption varies between 3.7 and 6.6 GJ/ t of clinker, with an average of 4.8 GJ/t in China and India. Most shaft-based technology plants are operated in developing countries, such as China and India. They are operated only in those countries where the lack of infrastructure and capital makes them viable. In China for instance the regional industrialisation policy and the prohibition of entities to own chain of plants make shaft kilns widespread. In India their share is 10%, while in China it represents over 80% of the installed capacities. Their usual size is between 20–200 t/day, and many of them are fully hand-operated. 2.3. Retrofitting options One of the most important mechanisms in the model is the possibility of retrofitting among technologies. Retrofitting in the model means the transformation of one technology into another, provided it is technologically feasible. Both investment and variable costs are considered to determine the economic worthiness of the option. The possible retrofitting options are presented in Table 3. The technological feasibility has been determined according to previous studies (European Commission, 2000a; Hendriks et al., 2000; CEMBUREAU, 1999a). In the case of shaft kilns an automatic development is built into the model, based on studies on the Chinese and Indian cement industries (Liu et al., 1995; Schumacher and Sathaye, 1999). These studies make an indepth analysis on the possible future development of shaft kilns. Their conclusion is that even in the longterm, shaft kilns will continue to give substantial part of the cement producing capacities of China and India, but with lower energy intensity. The assumed potential for improvement ranges from 10% to 30% in these studies. In the model a 15% fuel efficiency gain is assumed in a 30 years time horizon.

3. Model description CEMSIM6 is a simulation model, working with interconnected modules on cement demand (consumption) and supply (production), international trade, 6 The CEMSIM model is fully documented in Szabo´ et al. (2003). The model has been written in VENSIM 5.0 software. The data sources appear in Appendix A.

ARTICLE IN PRESS L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

76 Table 3 Feasible retrofit options in cement processes From/to Wet Semi-wet Semi-dry Dry long Dry preheater Dry pre-calciner

Wet

Semi-wet

Semi-dry

O

O O

Dry long

Dry preheater

Dry pre-calciner

O O O O

O O O O O

Consumption by region

International trade

Production by region Production by technology

COST/PRICE

Investment: -capacity planning -retrofitting

Population

GDP

POLES energy prices

Carbon value

Energy demand in the cement industry

Carbon emissions POLES final energy demand

Fig. 1. Schematic overview of the cement model.

capacity planning and energy demand. The model operates with behavioural equations calculating these variables, which are presented in this section in a nontechnical way. Fig. 1 describes the functioning of the cement model. The dotted frames indicate the interface variables linked to the POLES model. GDP and population are exogenous variables, and follow the same dynamic path as in the POLES model. The prices for the various fuel uses in the cement sector come from POLES. The carbon emissions and final energy demand of the cement sector is to be used as an input in the POLES model, once the two models are linked. The various sub-modules are introduced in the following part. The model covers 51 regions of the world. Countries of the European Union are individually modelled, while other countries are classified together (e.g. South American region, Middle East). In this study the 51 regions have been further aggregated into 10 regions in order to facilitate the presentation of the results (Table 4). The usual approach to compute cement consumption is to link cement demand to the income level. Yet given

the multi-regional character of CEMSIM and the relatively long time-horizon, up to the year 2030, a more flexible approach has been implemented. GDP growth and the GDP level together determine the income effect on cement consumption. This solution is similar to other studies dealing with long term commodity intensities (e.g. Gielen and Moriguchi, 2002), where the income elasticity is supplemented by an autonomous material efficiency improvement (usually regional- and income level-specific). In particular, the model computes cement consumption taking into account the evidence on the relationship between commodity intensity (defined as commodity consumption per unit of GDP) and per capita GDP. This function has been determined empirically for various materials and countries.7 Its shape consists of an inverse U-shaped curve (also called ‘Intensity of Use hypothesis’). The inverted U-shape can be explained in terms of 7 Related studies are Liu et al. (1995) for cement, Van Vuuren et al. (1999) for steel, and Mannaerts (2000) for steel, aluminium, paper and different chemical products.

ARTICLE IN PRESS L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

77

Table 4 Regional coverage Aggregated regions

Countries

North America Western Europe OECD Pacific CEU-FSU Latin America India China SE Asia Africa Middle East

US, Canada EU15, Rest of Western Europe (Cyprus, Malta, Norway, Switzerland) Japan, Australia, New Zealand Central and Eastern European countries, Former Soviet Union Central and South American countries India China Asian countries, except India and China African countries Middle East and Gulf countries, Turkey

Commodity Intensity

∋

Commodity intensity (Mton/M )

300 250 200 150 100 50 0 0

10

20 30 GDP per capita ( /cap)

40

∋

Fig. 2. Commodity intensity curve.

the superposition of three different trends (Van Vuuren et al., 1999) affecting the dynamic evolution of the commodity requirements (Fig. 2):



 

The different phases of the economic transition from agriculture to manufacturing and construction (as per capita GDP rises, the commodity intensity increases), and then to services (per capita GDP growth is accompanied by commodity intensity falls). The substitution away from the commodity, because of the use of alternative materials in the different applications. The result of technological development.

Developing countries such as China or India would be in the left-hand side of the curve (positive slope), whereas developed countries such as US or Germany would be located in the right-hand side (negative slope). Each country has its own consumption pattern, determined by two shape parameters estimated on 10–15 years consumption data. Current year cement consumption is determined by the last year consumption, income (GDP) and by the price change of cement. To arrive to the country-specific cement prices of the model, the total production cost is augmented by the transport costs on the imported cement, and a mark-up rate is used, as estimated in Martins and Scarpetta (1999).

Production of cement is determined in a two-level hierarchy system. First, the global cement market is allocated to the producing regions, where the shares are determined by the previous years production and the average production costs, taking into account the availability of capacities as well. Once the production is determined in a region, the second step is to allocate production to the available technologies using a dispatching method similar to that of the first step. From the regional cement consumption and production, the net trading positions of the different regions are straightforwardly calculated. Since international trade in cement accounts to 7% of world production this simplified approach gives reasonable results. Another characteristic of the international cement market is that countries who are net exporters have very limited imports, and net importers have limited exports of cement, as there is little scope for product differentiation. This also supports the approach of net trade calculations. Cement capacity data by technology is available in CEMBUREAU (2002) on a company basis, and have been aggregated to the regional level. New capacity can appear in the system in two ways. The first one is by building new utilities, and the second is by upgrading, transforming or expanding existing capacity (retrofitting). Retirement of capacities takes place according to the economic lifetime of the capacities. Capacity is kept during its entire lifetime, even if new and cheaper utilities are built. New capacity building is based on the expected market growth, which depends on the past developments of the markets8 (e.g. average production growth rate for the past ten years). If the expected production is higher than the available capacities (total minus retired capacities), new capacities will be built. The shares for 8 This adaptive expectations approach then does not follow the perfect foresight approach (e.g. base the investment decision on the discounted future overall costs). On technology dynamics see Kouvaritakis et al. (2000b) and Jacobsen (2000).

ARTICLE IN PRESS 78

L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

new capacities are set according to the total costs of these technologies, including investment costs. Investment costs are annualised through the economic life of the investment. The core of capacity planning is retrofitting, which takes into account both investment and variable costs. Only those technologically feasible options are allowed, which have been listed in Table 3. The demand for the different fuel types is determined by the variable fuel cost, and by a function of the investment cost related to the use of the fuel. Cement production is an energy-intensive industry. Most of the energy (over 85%) is consumed during the clinkerisation process, in the kiln through burning different types of fuels. The majority of the kiln technologies (except the shaft kilns) are flexible to use different fuel types (coal, coke, natural gas or oil), but storage facilities, fuel transport facilities and availability of fuels are decisive. Thus different regions have different portfolio of fuel used, mainly determined by the price and availability of different fuels.

4. Model results The model results are presented in two subsections. The tendencies concerning consumption, production and technology developments are presented in the business as usual (BAU) scenario. The second subsection studies the three policy-relevant scenarios concerning carbon emission trading. The alternative scenarios differ in the geographical coverage of the emission trading markets (EU15, EU27 and Annex B) and, as a consequence, in the resulting equilibrium carbon prices. The section finishes with a sensitivity analysis on the most important parameters of the model.

carbon emissions and, therefore, the carbon value is set to zero. GDP and population growth are exogenous variables, while energy prices come from the POLES model. The model computes endogenously a series of variables, as previously detailed. Certain additional hypotheses have been assumed to close the model. 4.1.1. Consumption, production and carbon emissions According to the BAU scenario cement consumption will grow at high rates on world level in the 2000–2030 period. On global level, the 1600 Mt of cement consumption in 2000 will increase almost twofold to 2880 Mt by 2030, implying an annual 2% growth rate. Fig. 3 represents the regional consumption of cement in 10-year intervals (1997 is given in the figure as the base year). The chart shows that most growth takes place in the developing regions. China and India experience the highest growth. In 2030 they will consume almost half of the world cement production. China reaches its maximum consumption by 2020 and then its consumption remains relatively stable till 2030. This is due to the assumed behaviour of cement consumption (the inverted U shape hypothesis). China, a country with very high growth in material intensity, will reach its peak earlier and then faces a decreasing

4.1. Reference scenario Table 5 introduces the BAU or reference scenario assumptions. It is assumed that there is not any tax on

Fig. 3. BAU cement consumption by region.

Table 5 Assumptions of the BAU scenario Time horizon

1997–2030

Carbon value

Set to zero

Exogenous variables, parameters

GDP (in PPP), population Primary fuel prices Price, income and activity parameters

Endogenous variables

Cement demand, production, net international trade Capacity building, retrofitting and retirement by type of technology Fuel mix in primary fuel demand Carbon emission

Closed assumptions

Autonomous technology trend for shaft kilns No fluidised cement technology, no alternative cement product (polymer) penetrate the market No change in clinker/cement ratio in cement production

ARTICLE IN PRESS L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

trend in total consumption, when its GDP and population growth rates gradually fall. This seems to be a plausible assumption, as after completing the majority of infrastructure developments, the maintenance of the infrastructure usually requires less amount of materials. Other developing regions, such as Latin America, Africa, Middle East and Rest of Asia also show high growth rates in cement consumption, but not as high as in China and India. Cement consumption in developed regions (e.g. North America, Western Europe and OECD Pacific) are stagnating. This means that these regions together with the Former Soviet Union and Eastern Europe would see their share in the world consumption fall. Production of cement shows a similar regional picture to that of consumption. However there are some significant differences. All regions show an increasing production trend till 2020. After 2020 only the Western Europe, Central Europe and the FSU, and the OECD Pacific regions show stabilised production. All developing regions experience an uninterrupted growing trend in cement production. In the case of China production does not show the U-trend seen in consumption. Production is still growing till 2030, but the production growth rate gradually slows down. This means that China will increase its production and export to other Asian regions, where production does not cover consumption. According to these projections China will be an even more dominant player not only in the Asian cement market, but also in the global market (Fig. 4). Energy consumption follows a dynamic path close to that of production. The decomposition of the changes into carbon and energy intensity figures reveals interesting trends in the sector. Energy intensities in all regions fall, showing significant and uninterrupted improvements in energy efficiency. However carbon intensities deteriorate, indicating that carbon-intensive fuels (coal and heavy fuel oil) become even more dominant in the future in the reference case. CO2 emissions basically depend on the production level, consumption of fossil fuel and the technology mix. The business as usual scenario projects an increase of CO2

79

emissions of 56% in 2030 (reaching more than 2100 million tons of CO2), while at the same time production increases 75% on world level. Therefore in the reference scenario significant CO2 abatement is expected to take place both through changes in the technology mix and energy efficiency improvements. 4.1.2. International trade of cement According to the model results, international cement trade will significantly grow, reaching 15% of total consumption in 2030 (compared to a share of 7% in 1997), with notable changes in the regional patterns (see Fig. 5).9 The CEU–FSU and the Western European regions will intensify their exports together with North America, which will become exporter by 2020. However the most important changes take place in China, which will shift in the 2010–2030 period from an importer status to the most important exporter between 2020 and 2030. This shift could be explained by the very intensive consumption growth, whereas capacity expansion is lagging behind up to the year 2010. The trends will completely change by 2030, when consumption growth slows down and the capacity is significantly expanded. The other developing regions will be cement importers. 4.1.3. Technology development The business as usual scenario results show the direction and the speed of the changes in the technology mix. The most advanced and less energy consuming dry pre-calciner technology rapidly increases its share by 2030, as depicted in Fig. 6. The dry preheater and the shaft technologies slightly increase their share until 2010, and then their shares remain almost unchanged. All the other technologies—including the wet technologies and the old dry technologies, which represented one quarter of the technology portfolio in 1997—are expected to disappear by the end of the period. They either are converted to semi-wet and semi-dry technologies or face closure. 4.2. Emission trading scenarios Emission trading is one of the three international market mechanisms foreseen by the Kyoto Protocol on climate change in order to meet the GHG reduction objectives. Emission trading is cost-effective because it guarantees the achievement of an emission target with the lowest overall abatement costs. In the absence of transaction costs the initial allocation of emission permits, while having potentially large distributional effects, does not change the final outcome, i.e. the emission reduction taking place in the most efficient way. The allocation of emission rights (such as grandfathering and auctioning, see e.g. Cramton and Kerr

Fig. 4. Cement production by regions.

9

The table is based on net trading positions.

ARTICLE IN PRESS 80

L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

Fig. 5. International trade in the cement industry.

Fig. 6. BAU technology shares.

(2002); European Commission (2003c)) determines the economic burden undergone by the participants. This sub-section studies the effects of three possible emission trading markets on the cement industry of each country: EU15, EU27 and Annex B (without US).10 It is assumed that no specific emission abatement measures are undertaken outside the market boundaries. In order to simulate the emission permit markets, the sectoral marginal abatement cost curves for the cement industry have been calculated by introducing carbon values (in the range from 0 to 250 h/t CO2) into the CEMSIM model. This follows the same procedure as that implemented in the POLES model (Kouvaritakis et al., 2000b). It should be noted that the marginal abatement costs estimated by the CEMSIM model measure the efficiency losses induced by the carbon tax. Computable general equilibrium models can provide with a more complete analysis of the consequences of carbon policies, in particular taking into account preexisting distortionary taxation and the terms of trade effects (e.g. Babiker et al., 2003). The introduction of a carbon value (tax) reduces the cement demand. As the carbon value increases, prices of the fuels used in the different processes rise according to 10 In this scenario Russia determines its ‘hot air’ sales maximising its revenues.

their carbon content, inducing fuel substitution. Coal is the most negatively affected by the introduction of the carbon values, while natural gas becomes the cheapest option. Electricity price changes as well at a different rate in each country, according to the fuel mix used for power generation. The energy price increase translates into different increments in production cost for the different technologies and regions. Technology change and retrofitting speed up according to these effects, as the fuel price will push forward to more fuel-efficient technologies, through new capacities or retrofitting. The emission trade market is assumed to be perfect (i.e. no market distortion, and zero transaction costs), and will be cleared at the price where all marginal cost curves are equalised. In this article the emission permit allocation to the cement industry is based on the so-called optimal allocation approach, which ensures that the national targets are achieved minimising the overall compliance costs. National sectoral targets for the cement industry in each member state are calculated assuming that the EU Burden Sharing Agreement11 (BSA) has to be fulfilled. The results from the POLES model have been used for this purpose. The procedure is illustrated in Fig. 7. POLES compute for each member state the permit price reached under national emission trading regimes, given the national BSA target. The permit prices under national emission trading regimes are given by the intersection between the national marginal abatement cost curves (MAC) and the BSA targets. The intersections between the sectoral MACs and the national permit prices give the emission allocation among all the industries included in the trading regime. Such an allocation ensures the fulfilment of the BSA targets. Once the POLES-based national permit prices are derived, the sectoral marginal abatement cost (MAC) curves from CEMSIM are used to compute the targets for the cement industry in each member state. The national targets and the respective national permit prices appear in Table 6. The EU15, EU27 and Annex B permit prices (signed with * in the last three rows of the table) indicate the equilibrium permit prices if full trade amongst the participating countries is allowed. The last three columns of the table present the national permit prices that would balance the national carbon markets if there were not European emission trading. The prices have been computed with the POLES model. In the case of EU27 and Annex B trade, the national permit prices in the candidate countries are 11 The EU greenhouse gas emission reduction target in the Kyoto Protocol is 8% by 2008–2012, from 1990 levels. This overall target has been distributed on a differentiated basis to individual Member States with an ‘EU burden sharing’ mechanism agreed upon by the Council of Ministers in June 1998. For a detailed analysis see e.g. Eyckmans et al. (2002).

ARTICLE IN PRESS L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

81

Fig. 7. Target setting in the policy scenarios.

Table 6 National BSA targets and permit prices from POLES Reduction target (%) Austria Belgium and Luxembourg Denmark Spain Finland France United Kingdom Greece Ireland Italy Netherlands Portugal Germany Sweden Cyprus Malta Hungary Poland Czech Republic Slovak Republic Bulgaria Romania Slovenia Estonia Lithuania Latvia Russia Ukraine Canada Japan Rest of Western Europe Australia and New Zealand Rest of Central Europe European Union—15 Trade European Union—27 Trade Annex B Trade


13
9
21 15 0 0
12.5 25 13
6.5
6 27
21 4 0 0
6
6
8
8
8
8
8
8
8
8 0 0
6
6
3.6

National permit price EU15

National permit price EU27

67 54

67 55

70 58

135 35 210 37 12 47 89 35 158 42 9 149

136 35 229 37 12 47 89 35 158 42 9 149 0 0 11 11 11 11 0 0 0 0 0 0

140 36 225 39 12 48 90 37 160 43 10 154 0 0 11 11 11 11 0 0 0 0 0 0 0 0 98 12 250 57


7.3
3
8
8
5

National permit price Annex B

0 28* 18* 15*

ARTICLE IN PRESS 82

L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

much lower than the prices in the EU15 member states, or even zero, as in the case of some Eastern European12 and Former Soviet Union states, due to their declining economic output in the transition period. Their implied declining carbon emissions put these countries in the position to sell carbon credits, without real abatement. At the same time zero permit price means that the country does not have to constrain the emissions of its industries, as they are able to fulfil their Kyoto commitment with their projected BAU emission. Using the marginal abatement curves the compliance costs of fulfilling the national targets are calculated for each region participating in the emission trading scheme (see Babiker et al., 2002; Gusbin et al. 1999). The difference between the compliance cost of the trade and non-trade case gives the benefit arising from the international emission trading. The gains for the participating nations come from two sources. Those countries having high national permit prices (e.g. Netherlands, Denmark, Finland) are better off because buying permits abroad allows them to reduce their expensive domestic abatement effort (avoiding the high implied costs represented by the peak of their MAC curves). Countries characterised by cheap abatement options can increase their abatement effort, and sell the additional permits on the market. These benefits have been quantified for the three scenarios. The results appear in Table 7. The second column shows the estimated 1990 CO2 emissions of the sectors, while the third one is the BAU projection by the CEMSIM model. The last six columns show the three alternative scenario results, the reduction targets (effort) compared to the BAU case, and the quantified benefits for the trading countries. For the non-participating countries these benefits and targets logically cannot be calculated. The last three rows summarise the impact on the entire emission trading markets. The EU15-wide emission trading scheme would reduce the compliance costs of fulfilling the Kyoto target by 50.8 Mh in the cement sector (from 109 Mh in the national or isolated markets case to 58 Mh in the trading case). The cement sector would be a net buyer of permits at the EU-wide permit price, 28 h/t CO2. The Netherlands (14.71 Mh), Finland (11.36 Mh) and Sweden (8.23 Mh) would benefit most from the emission trading, as they have the highest compliance costs in the non-trading case. The cement industries of Germany and United Kingdom would receive revenues in the trading market. The overall reduction target of the EU15 is 2.4%, defined with respect to the 2010 BAU emissions. At the 12 The Czech Republic, Hungary, Poland and the Slovak Republic have the same permit price, as the current version of the POLES model groups them into the same region.

level of the individual countries the reduction targets show high variety, ranging between 0.8% and 17.9%. In general large emitters, such as Italy, Germany, Spain and France have lower emission targets, while northern countries face more stringent targets. The EU27-wide emission trading scheme would increase the benefits arising from trade to 67 Mh (from 111 Mh in the national market case to 43 Mh in the trading case). Only a smaller part of this benefit (2 Mh) comes from the direct emission reduction activities of the accession countries.13 The reduction target compared to the 2010 BAU varies between 0% and 19%, but on average the reduction requirement remains at a relatively low level (1.9%). The permission price compared to the EU15 trading case would fall to 18 h, which implies that most of the EU15 countries would increase their benefits from the widening of the emission trading market. Only Germany and the Great Britain would face lower benefit levels, as the income from selling their excess allowance would drop compared to the EU15-wide trade. Benefits would mainly be concentrated in four northern countries (Finland, Netherlands, Sweden and Denmark), as in the EU15 trade case. The Annex B emission-scheme would further increase the benefits of international carbon trade for the cement sector to 99 Mh (compliance cost reduces from 148 Mh in the national market case to 49 Mh in the trading case). The equilibrium price further reduces to h15, according to the POLES model simulations.14 Reduction targets compared to the 2010 BAU case varies between 0 and 18.5%, and the Annex B average stands at 2.5%. Europe (EU27) and ROWE would be the winners as most of the benefits are concentrated in these regions. In those countries where the equilibrium permit price is zero (e.g. Russia and Ukraine) the table shows the benefits without the income from ‘hot air’ (see footnote 13). The Russian Federation and Ukraine (and many other transition countries as well) have a stabilisation target for the first Kyoto period, but due to their economic decline, their emissions in 2010 will be much below those of 1990. This implies that these economies at first will be able to sell large amounts of rights, without indeed reducing emissions, and benefiting from potential financial transfers.

13 In order to make the alternatives comparable, revenues from ‘hot air’ are not included in the benefit calculation. This would also require an additional assumption on the allocation rule of the existing ‘hot air’ to the different sectors. The sector’s revenues from the ‘hot air’ sales could reach up to 120 Mh in the case of the Former Soviet Union region (mainly Russia), and 72 Mh for Central and Eastern Europe. 14 With the participation of the US the equilibrium carbon price would be h18, assuming that Russia sales all its available hot air. This excess supply is overbalanced by the US demand for permits.

ARTICLE IN PRESS L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

83

Table 7 Impacts of alternative emission trading schemes on the cement sector 1990 emissions (Mt)

Austria Belgiuma Denmark Spain Finland France United Kingdom Greece Ireland Italy Netherlands Portugal Germany Sweden Cyprus Malta Hungary Poland Czech Republic Slovak Republic Bulgaria Romania Slovenia Estonia Lithuania Latvia Russia Ukraine Canada Japan Rest of Western Europe Australia and New Zealand Rest of Central Europe EU15 Trade EU27 Trade Annex B Trade a

2010 BAU emissions

EU15 trade

EU27 trade

Annex B trade

Reduction target compared to BAU 2010 (%)

Benefits (Mh)

Reduction target compared to BAU 2010 (%)

Benefits (Mh)

Reduction target compared to BAU 2010 (%)

Benefits (Mh)

3.51 5.42 1.19 22.35 1.20 16.53 10.54

2.91 8.51 1.55 23.40 1.01 13.49 10.79


4.7
2.9
7.9
2.2
17.9
2.2
0.8

1.51 1.51 4.57 0.36 11.36 0.32 0.89


4.7
2.9
7.7
2.2
19.0
2.2
0.8

2.34 3.01 5.38 2.09 14.69 1.39 0.12


4.2
2.8
8.2
2.0
18.5
2.0
0.6

2.34 3.27 6.46 2.01 15.34 1.50 0.07

9.75 1.17 30.80 2.67 5.22 23.17 1.78 0.82 0.00 3.19 13.99 6.31

9.44 1.83 29.51 2.67 6.95 26.28 1.98 1.31 0.00 3.64 12.46 3.99


2.2
7.9
2.1
12.5
3.2
0.5
8.8

0.81 2.81 0.42 14.71 0.51 2.83 8.23


2.2
7.8
2.1
12.1
3.2
0.5
8.6 0.0 0.0
0.7
0.6
0.7

1.85 3.74 2.45 16.60 1.48 0.63 9.34 0.13 0.00 0.06 0.17 0.06


1.8
6.7
1.9
10.8
2.8
0.5
7.8 0.0 0.0
0.6
0.4
0.6

1.62 3.39 2.71 16.68 1.39 0.38 9.05 0.13 0.00 0.04 0.10 0.04

2.05

2.98


0.4

0.03


0.3

0.02

3.94 9.47 1.03 0.73 0.59 2.68 65.50 21.28 10.24 53.22 4.71

4.70 9.52 0.87 1.67 0.55 3.08 58.62 21.61 10.46 70.25 4.07

0.0 0.0 0.0 0.0 0.0 0.0

0.26 0.99 0.10 0.06 0.03 0.00

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
1.5
0.3
5.6

0.12 0.36 0.09 0.00 0.01 0.07 1.22 0.22 5.29 0.21 23.22

6.22

5.72


1.5

1.14

6.93

4.28

0.0

0.49


2.5

98.98


2.4

50.83
1.9

411.19

439.88

67.00

Together with Luxemburg.

4.3. Carbon leakage A generally cited disadvantage of emission trading schemes is that part of the production of the countries subject to the emission constraints might be relocated to countries not participating in the trading scheme, a

phenomenon known as carbon leakage. Non-participating regions might then increase their market shares, as their production is not penalised by the carbon price. The environmental effectiveness of the regional emission trading market diminishes when this world perspective is taken into account. The CEMSIM model can quantify

ARTICLE IN PRESS 84

L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

Appendix B reports on the implementation of the sensitivity analysis. Given the 33 year time horizon and the wide range in which the parameters have been allowed to change, the sensitivity analysis results suggest that the model BAU is relatively robust. The results also show that the capacity planning module is the most sensitive part of the model. Thus in the further development of the CEMSIM model this part would require more scrutiny.

Fig. 8. Production leakage as a function of the EU15 emission permit price.

this kind of effects because it is a global model.15 In the following the EU15 case is studied. Given that this market has the smallest geographical coverage, the carbon leakage effect is expected to be the highest in magnitude among the three considered scenarios. When more countries are involved in the trading scheme, the leakage effect would logically decrease. Fig. 8 shows the size of the production change with respect to the BAU scenario resulting from the introduction of permit prices in the range of 0–50 h/t CO2. For example, at the permit price of 40 h/t CO2 the EU15 will produce 5 Mt less than in the BAU case (around 3.5% of its total production), while the other regions will increase their production, however to a less extent. The ratio between the production increase in other region and the reduced production in the EU15 is 29% at this price. The last bars (World-EU15) represent the overall production leakage effect for the regions other than the EU15. This leakage effect rises as the permit price increases. 4.4. Sensitivity analysis The robustness of the results to crucial assumptions and parameters has been assessed with a detailed sensitivity analysis. A series of model runs has been performed with different values of some selected parameters (mainly price (costs) and income elasticities). The analysis has been carried out following a two-step procedure. In a first stage, five model parameters have been selected among a set of 12, using a sensitivity index (introduced later) that represents the sensitivity of the model results to these parameters. In the second stage, the Latin hypercube method has been applied to check the impact of the selected parameters when all do change at the same time. 500 model runs have been made in this second stage. 15 As the carbon value is introduced in the countries participating in the emission trading market, the international trade flows of cement of all the 51 CEMSIM regions do change, and are accounted for in the model results.

5. Conclusions This article has presented a recursive simulation model of the cement industry, the CEMSIM model. The model has been used to study the most important trends in the world cement market concerning production, technology development, energy consumption and carbon emissions in the 2000–2030 period. The article also analyses the potential impact on the cement sector derived from the implementation of new market mechanisms aimed to fulfil the carbon reduction targets set up by the Kyoto Protocol on climate change. In particular, the benefits of three potential international emission trading markets have been calculated (EU15, EU27 and Annex B). In the reference scenario a sharp increase in world cement production is expected. The 1550 Mt of cement production in 1997 is projected to reach 2800 Mt by 2030. This increase is due to the intensive growth of cement consumption in developing regions, mainly China, South-East Asia and India, tied to their intensive industrialisation processes. On the contrary, developed regions, such as North America, Europe and OECD Pacific, have stabilising or declining consumption patterns, while their cement production is expected to rise slightly. International trade in cement shows an uninterrupted growing trend. It is projected to increase from 112 Mt in 1997 to 450 Mt in 2030, which translates into a rise in the share of international trade in consumption from 7% to 16%. Developed regions are the main exporters, while most of the developing regions import an increasing share of their consumption. China is the only exception, a country projected to be the largest exporter by 2020. Concerning the evolution of the technology portfolio the most advanced dry-preheater and dry-pre-calciner technologies are expected to become the most widespread technologies by 2030. Technologies with higher energy consumption, such as the wet long and dry long kilns will be driven out from the market or will be retrofitted. Global CO2 emissions from the cement industry will increase by more than 50% in the BAU scenario, reaching 2100 Mt by 2030. The growth of carbon

ARTICLE IN PRESS L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

emissions is less intensive than the growth in cement production due to the shift to cleaner technologies and the improvements in energy efficiency. But the results also suggest that the sector will continue to increase the shares of carbon-intensive fuels in both the reference and carbon trading cases. The impacts of three different CO2 emission trading schemes have been numerically calculated with the model. Under the EU15-wide emission trading scheme the costs of fulfilling the Kyoto Protocol targets in the cement sector would be reduced by 50 Mh. This benefit is quite concentrated in some northern European countries. At the equilibrium price of 28 h/t CO2 (projected by the POLES model) the cement industries of most of the EU-15 countries would be permit buyers. If the trading scheme is enlarged to the EU27, the permit price would fall (to 18 h/t CO2) and benefits for the participating countries would increase to 67 Mh. Accession countries would mainly benefit from their ‘hot air’, but their projected growing cement consumption reduces their trading potential by 2010. An Annex B-wide emission trading would entail a 15 h/t CO2 permit price as well, with additional benefits for the participants, estimated to amount to 99 Mh. These numbers are in accordance with the theory, as increasing the ‘pool’ size of participating countries should decrease overall compliance costs and raise benefits. On the other hand, according to the model results the sub-global coverage of emission trading does not lead to very high carbon leakage effects till 2010 in the considered permit price range (0–50 h/t CO2). However with increasing permit price beyond this range the leakage effect might intensify. Finally, the quantitative results of this study do not intend to be forecasts or predictions about the future evolution of the cement sector. The CEMSIM model provides with numerical results, taking into account the current available data and the relationships between the model variables. The model results are therefore to be interpreted in the particular proposed analytical framework. Undoubtedly, there remain significant data problems and poorly understood key relationships. The ultimate purpose of this research has been to provide with useful insights for policymakers.

Acknowledgements A preliminary version of this article was presented at the EAERE conference held in Bilbao, June 2003. The authors would like to acknowledge the comments and suggestions received from Santiago Rubio (University of Valencia), Peter Russ (IPTS), and an anonymous referee.

85

Appendix A. Sources of databases CEMBUREAU, 1999b. World Statistical Review no. 18 (1913–1995) and no. 19–20 (1994–1995). CEMBUREAU, 2002. World Cement Directory 1996 & 2002. Department of Trade and Industry (UK), 2001. Construction Statistics Annual (http://www.dti.gov.uk/ construction/stats/stats2001/pdf/constat2001.pdf). European Environmental Agency, 2002. Greenhouse gas emission trends in Europe 1990–2000. ISBN 92-9167516-4. International Energy Agency (IEA), 2002a. Energy Statistics of OECD Countries. International Energy Agency (IEA), 2002b. Energy Statistics of Non-OECD Countries. International Cement Review (ICR), 1998. The Global Cement Report (3rd Edition). United Nations Framework Convention on Climate Change (UNFCCC), 2003. Database on GHG gases (http://unfccc.int/). United Nations Population Division (UN), 2003. World Population Prospects: The 2002 Revision Population Database (http://esa.un.org/unpp).

Appendix B. Sensitivity analysis This annex details the sensitivity analysis carried out in the CEMSIM model. The parameters included in the analysis are quite different in their character and calculation. Some of them are classical elasticity parameters (based on cost or prices), while others are auxiliary variables. Table 8 gives the base values of the parameters (those used in the reference scenario), together with the test values, used as boundaries for the sensitivity runs. Some parameters are allowed to change in the range of 750% compared to the reference value. For the income elasticity parameters a smaller range has been assumed, as they are econometrically estimated values on the past 15-year data. For the exogenous variables (per capita GDP, population data) the range is set according to the UN and EIA estimates.16 The sensitivity index is defined in the following expression.17 vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi !2 u base Xu X test t itr
X itr : I ir ¼ X base t itr 16 The United Nations World Population Prospect 2002 revised database (UN, 2003) has been used for population. The EIA Annual Energy Outlook GDP estimates (Energy Information Administration (EIA), 2002) have been considered as the basis for the range setting high and low economic growth cases. 17 The index is constructed on the basis of Nordhaus (1994).

ARTICLE IN PRESS L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

86 Table 8 Parameter values and sensitivity ranges Parameter

Abbreviation

Base value

Range

Composite income elasticities Cost share elasticity Discount rate Price elasticity Market allocation parameter Retrofitting substitution parameter Overall retrofit limit parameter Retrofit limit parameter by technology New capacity limit parameter Per capita GDP Population

ACCEM, BCCEM CSE DISR PRE MAP RSP RLPO RLPT NCLP GDPPOP POPULATION

Country-specific 2 0.1
0.2 2 Technology-specific (
0.1 or
0.01) 0.25 Between 0–0.1 1.5 Country-specific Country-specific

Reference value 1.0–3.0 0.05–0.15
0.1 to
0.3 1–3 Reference value 0.125–0.375 Reference value 0.75–3 Reference value Reference value

The term Xtest itr is the value of the variable Xi at time t for region r in the test run. The term Xbase itr is the value of the variable Xi at time t for region r in the base run. The index represents the standard deviation of the test runs with respect to the base run for the different time periods, in particular for 2010, 2020 and 2030. Both differences for the upper and lower test values (given in Table 8) are calculated, as it is expected that some of the parameters do not behave symmetrically. The effects of the sensitivity parameters are examined on the following four result variables, which could be considered as the most relevant variables of the model: consumption, production, carbon emission, and total capacity. The overall sensitivity index I is defined in the following expression: 1XX I¼ oir I ir ; 4 i r where oir represents the weight of variable i and region r. The four result variables have the same weight, 14. The weights oir are derived taking into account the regional weights of the variable.18 The formula computing those weights is the following: X ir oir ¼ P : X ir

725%

750% 750% 70.6% (on yearly basis) 70.25% (on yearly basis)

Table 9 Sensitivity results for the world region

Expected value Consumption Production Emission Capacity

2010

2020

2030

2159 2159 1678 2242

2604 2825 2035 3166

2853 3094 2165 3552

146 146 101 157

303 870 505 1092

464 897 539 1087

Standard deviation Consumption Production Emission Capacity Coefficient of variation Consumption Production Emission Capacity BAU value Consumption Production Emission Capacity

0.07 0.07 0.06 0.07

2169 2169 1691 2223

(Expected–BAU)/BAU Consumption
0.46% Production
0.46% Emission
0.77% Capacity 0.85%

0.12 0.31 0.25 0.34 2613 2612 1951 2702
0.34% 8.15% 4.31% 17.17%

0.16 0.29 0.25 0.31 2884 2883 2105 3002
1.07% 7.32% 2.85% 18.32%

r

The five parameters with the highest sensitivity index are the capacity limit parameter (NLCP), the income elasticities (ACCEM, BCCEM), per capita GDP (GDPPOP), and population (POPULATION). With regard to the Latin hypercube analysis, the variation ranges for the five selected parameters have been determined according to the two extreme values, the base value, and two mid-points between the 18 For comparison, an alternative index has been also calculated when all regions get the same weight, producing very similar results. In the case of the total capacity variable, the capacity shares by technology are more important than the shares by regions. Hence in this case the technology shares have been used in the calculation of the index.

Units in Mt, emission in Mt of CO2.

extremes and the base value. In this way the number of runs is 55=3125. The number of runs has actually been limited to 500 in order to keep the analysis within reasonable computable limits. The Latin hypercube method ensures that the full range of variation of each parameter is explored, given the large number of simulations run. Table 9 summarises the results of this analysis. The table presents the expected values and the standard deviations of the result variables. The coefficient of variation shows the average variation values (defined as the ratio of the standard deviation to the expected

ARTICLE IN PRESS L. Szabo´ et al. / Energy Policy 34 (2006) 72–87

value). The coefficient of variation shows its peak in 2020. While the standard deviation values are increasing, the expected values show much smaller increases, resulting in smaller coefficients. Furthermore, another indicator of the table measures the deviation of the expected value with respect to the BAU value ((Expected Value–BAU)/BAU). The expected value of the sensitivity runs and the BAU scenario values are rather close to each other, with the exception of the capacity variable. References Babiker, H.M., Reilly, J.M., Viguier, L., 2002. Is international emissions trading always beneficial? MIT Joint Program on the Science and Policy of Global Change Report No. 93. Babiker, H.M., Criqui, P., Ellermann, D., Reilly, J., Viguier, L., 2003. Assessing the impact of carbon tax differentiation in the European Union. Environmental Modeling and Assessment 8 (3), 187–197. Capros, P., Kouvaritakis, N., Mantzos, L., 2001. Economic evaluation of sectoral emission reduction objectives for climate change: TopDown analysis of Greenhouse Gas Emission possibilities in the EU’ (PRIMES). Contribution to a Study for DG Environment, European Commission. CEMBUREAU, 1999a. Best available techniques for the cement industry. A contribution from the European Cement Industry to the exchange of information and preparation of IPPC BAT Reference Document for the cement industry (www.cembureau.com). Cramton, P., Kerr, S., 2002. Tradeable carbon permit auctions: How and why to auction not grandfather. Energy Policy 30, 333–345. Energy Information Administration (EIA), 2002. Annual energy outlook 2002 (http://www.eia.doe.gov/analysis/2001anal01.html). European Commission, 1996. POLES 2.2. European Commission, DGXII, JOULE II Programme, Science Research Development. Ref EUR 17356 EN. European Commission, 2000a. Best available techniques reference document in the cement and lime manufacturing industries. European Integrated Pollution Prevention and Control Bureau (EIPPCB), Institute for Prospective Technological Studies, Sevilla, Spain European Commission, 2000b. Green paper on Greenhouse Gas Emission trading within the European Union. COM 2000-87. European Commission, 2003a. Directive of the European Parliament and of the Council establishing a scheme for greenhouse gas emission allowance trading within the Community and amending Council directive 96/61/EC (Directive 2003/87/EC). European Commission, 2003b. Common position of the Council on the adoption of a Directive of the European Union and of the Council establishing a scheme for greenhouse gas emission allowance trading within the Community and amending Council Directive 96/61/EC. Communication from the Commission to the European Parliament. SEC 2003-364 Final. European Commission, 2003c. The EU emission Trading Scheme: how to develop a National Allocation Plan. Non Paper Second meeting of working 3 Monitoring Mechanism Committee, April, 2003. Eyckmans, J., Cornillie, J., Regemorter, D., 2002. Efficiency and equity in the EU Burden Sharing Agreement. ETE Working Paper No. 2000-02, Katholieke Universiteit Leuven.

87

Gielen, D.J., Moriguchi, Y, 2002. CO2 in the iron and steel industry: an analysis of Japanese emission reduction potentials. Energy Policy 30, 849–865. Gusbin, D., Klaassen, G., Kouvaritakis, N., 1999. Cost of a ceiling on Kyoto flexibility. Energy Policy 27, 833–844. Hidalgo, I., Szabo´, L., Calleja, I., Ciscar, J.C., Russ, P., Soria, A., 2003. Energy consumption and CO2 emissions from the world iron and steel industry. Institute for Prospective Technological Studies, Joint Research Centre, Report EUR 20686 EN, Seville, Spain. Hendriks, C.A., Worrell, E., Jager, D., Blok, K., Reimer, P., 2000. Emission reduction of Greenhouse Gases from the cement industry. Greenhouse Gas Control Technologies Conference Paper. (www.ieagreen.org). International Energy Agency (IEA), 1999. The reduction of greenhouse gas emission from the cement industry. IEA Greenhouse Gas R & D Programme Report No. PH3/7. Jacobsen, H.K., 2000. Technology diffusion in energy-economy models: the case of Danish Vintage Models. The Energy Journal 21, 43–71. Kouvaritakis, N., Criqui, P., Thonet, C., 2000b. World post-Kyoto scenarios: benefits from accelerated technology progress. International Journal of Global Energy Issues 14 (1–4), 184–203. Liu, F., Ross, M., Wang, S., 1995. Energy efficiency of China’s cement industry. Energy 20 (7), 669–689. Mannaerts, H., 2000. STREAM: substance throughput related to economic activity model: a partial equilibrium model for material flows in the economy. CPB Research Memorandum No. 165, The Hague, ISBN 90 5833 042 7. Martins, O. J., Scarpetta, S., 1999. The levels and cyclical behaviour of mark-ups across countries and market structures. OECD Economics Department Working Papers No. 213. Nordhaus, W.D., 1994. Managing the Global Commons: The Economics of Climate Change. The MIT Press, Cambridge, MA. Rotman, J., Kelmanzky, D., 1997. Determinantes del consumo del Cemento en la Argentina. Paper presented at the Annual AAEP Conference (http://www.aaep.org.ar/espa/anales/pdf/rotman_kelmanzky.pdf). Schumacher, K., Sathaye, J., 1999. India’s cement industry: productivity, energy efficiency and carbon emissions. Energy Analysis Program: Environmental Energy Technologies Division, Lawrence Berkeley National Laboratory. Szabo´, L., Hidalgo, I., Ciscar, J.C., Soria, A., Russ, P., 2003. Energy consumption and CO2 emissions from the world cement industry. Institute for Prospective Technological Studies, Joint Research Centre, Report EUR 20769 EN (http://www.jrc.es/home/publications/publication.cfm?pub=1108). Van Vuuren, D.P., Strengers, B.J., De Vries, H.J.M., 1999. Long-term perspectives on world metal use—a system dynamics model. Resources Policy 25, 239–255. Worrell, E., Martin, N., Pryce, L., 2000. Potentials for energy efficiency improvements in the US cement industry. Energy 25, 1189–1214.

Further reading Kouvaritakis, N., Soria, A., Isoard, S., 2000. Modelling energy technology dinamics: methodology for adaptive expectations models with learning by doing and learning by searching. International Journal of Global Energy Issues 14 (1–4), 104–115.